Behavioral Finance
Classical finance has a beautiful model of how markets work: rational agents process all available information, price assets correctly, and markets are efficient. The model is elegant, internally consistent, and wrong about humans in ways that cost people real money.
Behavioral finance starts from the other end. Instead of assuming people are rational and deriving what should happen, it observes what people actually do and tries to understand why. The findings aren’t flattering.
The biases that matter
Kahneman and Tversky’s work in the 1970s and 80s laid the foundation. They showed that the ways people deviate from rationality aren’t random - they’re systematic, predictable, and rooted in the same cognitive shortcuts that bounded rationality describes. These aren’t bugs in rare edge cases. They’re default behaviors that show up in every portfolio review, every trading decision, every market cycle.
Loss aversion is probably the most consequential. Kahneman and Tversky found that losses feel roughly twice as painful as equivalent gains feel good. Losing 100 feels nice. This asymmetry warps behavior in predictable ways. Investors hold losing positions too long, hoping they’ll recover, because selling would mean accepting the loss as real. They sell winners too early, locking in gains before they can evaporate. The result is a portfolio of losers you’re emotionally attached to and winners you let go of prematurely - the exact opposite of what any rational strategy would produce.
I’ve felt this myself. Not in stock picking (I mostly index), but in projects. Continuing to invest time in something that clearly isn’t working because abandoning it would mean admitting the time already spent was wasted. That’s the sunk cost fallacy, and it’s loss aversion wearing a different hat. The loss of admitting failure feels worse than the ongoing cost of pretending things might turn around.
Overconfidence is the bias that’s hardest to defend against because by definition, you don’t think you have it. Studies consistently show that active traders underperform index funds, and the ones who trade most frequently perform worst. The pattern is clear: people overestimate their ability to pick stocks, time markets, and interpret information better than the crowd. Trading costs (fees, spreads, taxes) compound the damage, but the core problem is the belief that you know something the market doesn’t.
Overconfidence doesn’t just show up in trading. Every planning estimate that misses by 3x is overconfidence. Every project that was going to take “a couple of weeks.” Every architecture decision made without enough research because you were sure you knew the right answer. The planning fallacy is overconfidence applied to timelines.
Anchoring is subtle and everywhere. The first number you encounter in a negotiation, analysis, or evaluation becomes a reference point that all subsequent numbers are judged against. A stock that was at 50 feels “cheap” - even if 800k that drops to 600k.
Anchoring is why initial offers matter so much in negotiations. It’s why “was 99” works even when the thing was never actually worth $199. And it’s insidious in technical decisions too - if someone suggests a timeline of 6 months, all subsequent estimates will cluster around that number regardless of the actual complexity.
Herd behavior creates the big, dramatic failures. Individual loss aversion, overconfidence, and anchoring create personal mistakes. Herding creates systemic ones. When everyone is buying, buying feels safe and selling feels risky. When everyone is selling, selling feels prudent and buying feels reckless. The information in the crowd’s behavior gets amplified and distorted until the market reaches a point where prices are completely disconnected from underlying value.
Every bubble has the same shape. Tulips, dot-com, housing, crypto. The asset changes but the behavioral dynamics are identical. Early participants make real money. Later participants see those returns and pile in. Prices rise because people are buying, and people are buying because prices are rising. Everyone has a theory for why this time is different. I was around for crypto in 2017 and again in 2021, and the conversations were the same both times - smart people with sophisticated-sounding narratives explaining why the fundamentals supported the price, right up until they didn’t. Nobody wants to be the first to leave the party. Then the music stops, and the same herding dynamics that inflated the bubble collapse it - faster, because panic is a stronger motivator than greed.
Recency bias amplifies everything else. Recent experience dominates over base rates. A market that’s been rising for three years feels like it will keep rising. A market that just crashed feels dangerous even if valuations are now attractive. Recency bias is why people invest the most at market peaks (recent returns are great!) and the least at market bottoms (recent returns are terrible!). It’s systematically buying high and selling low.
Why knowing doesn’t fix it
What makes behavioral finance humbling rather than just intellectually interesting: knowing about these biases doesn’t make you immune to them. I know about loss aversion. I still feel the pull to hold onto losing positions. I know about anchoring. The first number I hear still warps my subsequent thinking.
This is because the biases aren’t errors in reasoning that more information can fix. They’re features of how human cognition works. They’re the heuristics that let us make decisions quickly in a complex world. They work well in most contexts - loss aversion keeps you from taking reckless risks, herding keeps you aligned with your community, recency bias keeps you responsive to changing conditions. They fail in financial markets because markets are an adversarial environment that actively exploits these tendencies.
The practical question isn’t “how do I become rational?” It’s “how do I build systems that protect me from my own predictable irrationality?”
Automatic investing bypasses loss aversion and recency bias by removing the decision point entirely. Dollar-cost averaging into an index fund means you buy regardless of how the market feels. You buy in bubbles (buying less, since prices are high) and in crashes (buying more, since prices are low). Not because you’re smart enough to time it, but because the system doesn’t let you opt out based on emotion.
Rebalancing rules force you to sell high and buy low mechanically. If your target is 80/20 stocks to bonds and stocks surge, rebalancing sells some stocks (locking in gains you’d otherwise hold too long) and buys bonds. If stocks crash, it buys more stocks (at exactly the moment your instincts scream to sell). The rule counteracts the bias.
Checklists before trading introduce friction between impulse and action. Write down your thesis, your exit criteria, and what evidence would prove you wrong, before you buy. Then review it when you feel like selling. It doesn’t eliminate bias, but it slows down the process enough to engage deliberate thinking instead of reflexive emotion.
The meta-lesson
Behavioral finance is really a specific application of a broader insight: humans are predictably irrational in ways that can be understood and partially mitigated. The same biases that show up in financial markets show up in engineering decisions, in game-theoretic situations, in how we evaluate our own work, in how we plan projects.
The useful frame isn’t “I should be more rational.” It’s “I should assume I’m irrational in predictable ways and design systems accordingly.” That’s the same argument as desired state systems - you can’t prevent drift, so you build correction mechanisms. You can’t prevent cognitive bias, so you build processes that counteract it.
The goal isn’t to be perfectly rational. It’s to be less systematically wrong than you’d be without guardrails.